WO2021045018A1 - Design support system, design support method, and program - Google Patents

Abstract

A design support system (1) comprising a variable extraction means and a setting means. In the design support system (1), the variable extraction means extracts variables that stipulate design constraints, from validated design data. The setting means sets the design constraints by statistical processing of the frequency distribution of variables that have been extracted by the variable extraction means or by using machine learning to analyze same.

Description

Design support system, design support method and program

This disclosure relates to design support systems, design support methods and programs.

Design drawings showing the mechanism of mechanical devices, circuit layout of semiconductor devices, etc. include design information such as the shape, size, and coordinates of parts such as mechanical parts and circuit elements. Each component must be arranged in compliance with predetermined design constraints so that positional overlap, operational interference, etc. do not occur. Therefore, it is checked whether or not the design items comply with the design constraints.

However, the design constraints that give the distance between parts are generally determined with a margin. Therefore, the distance between parts tends to be large. Therefore, when the device is designed by adopting the conventional design constraint, the designed device may be larger than the predetermined size or the cost may be high. Therefore, within the range where the final performance can be expected to satisfy the target value, the design constraint may be exceptionally relaxed, the design drawing may be passed, and the design may be established.

Patent Document 1 discloses a circuit design support device that supports work for exceptionally relaxing design restrictions. This circuit design support device includes a pseudo error registration file. If the user determines that the design item that the circuit design support device has determined to violate the design constraint is not a true error, the user registers the error in the pseudo error registration file. When the design item determined to be an error according to the design constraint is registered in the pseudo error registration file, the circuit design support device cancels the error and passes the error.

Japanese Unexamined Patent Publication No. 4-36866

In the technology described in this patent document, there is no standard for determining whether a design item determined to be an error based on a design specification is a true error or a false error. For this reason, the personality appears in the determination of the user. In addition, only when the same design item as the design item determined to violate the design rules is registered in the pseudo error registration file, it is automatically determined that the error is not a true error. In other cases, the user's identification is requested. Therefore, the burden on the user is large.

This disclosure was made in view of the above circumstances, and an object of the present disclosure is to make it easy to set design constraints.

In order to solve the problems described above, the design support system of the present disclosure is
Variable extraction means for extracting variables that determine design constraints from verified design data,
A setting means for setting the design constraint by analyzing the frequency distribution of the variable extracted by the variable extraction means by using statistical processing or machine learning.
To be equipped.

According to the present disclosure, the design constraints are set by extracting variables that determine design constraints from the verified design data and analyzing the frequency distribution of the extracted variables using statistical processing or machine learning. Therefore, design constraints can be easily set.

Configuration diagram of the design support system according to the embodiment of the present disclosure The figure which illustrates the design data to be verified by the design support system shown in FIG. The figure which illustrates the radius of a circuit element in the design drawing shown in FIG. 2, the distance between circuit elements, and the external distance between circuit elements. The figure which illustrates the existence probability of the external distance in the design data stored in the past design DB shown in FIG. The figure which illustrates the discrete existence probability of the external distance and the cumulative existence probability in the design data stored in the past design DB shown in FIG. Flow chart of threshold value calculation process executed by the threshold value calculation device shown in FIG. Flow chart of the design verification process executed by the design verification device shown in FIG. The figure which shows another example of the existence probability of the external distance in the design data stored in the past design DB shown in FIG. The figure which shows the discrete existence probability of the external distance and another example of the cumulative existence probability in the design data stored in the past design DB shown in FIG. The figure which shows the example of the variable which concerns on Embodiment 2 of this disclosure and its existence probability. The figure for demonstrating the process of the design support system which concerns on Embodiment 3 of this disclosure. The figure which shows the process in the design support system which concerns on Embodiment 4 of this disclosure. The figure which shows the structure of the data stored in the threshold value DB of the design support system which concerns on Embodiment 5 of this disclosure. The figure which shows the example of the data stored in the past design DB which concerns on Embodiment 6 of this disclosure. The figure which shows the example of the design data and the variable which concerns on Embodiment 6. Flow chart of threshold value calculation process executed by the threshold value calculation device according to the sixth embodiment The figure which shows the example of the frequency table which concerns on Embodiment 6. The figure which shows the example of the design constraint table which concerns on Embodiment 6. The figure which shows the example of the frequency distribution of pass and fail which concerns on Embodiment 6. The figure which shows the example of the correspondence relation between the variable and the design constraint which concerns on Embodiment 7 of this disclosure. Flowchart of verification process executed by the design verification device according to the eighth embodiment of the present disclosure. The figure which shows the example of the verification result output file which concerns on Embodiment 8. FIG. 1 is a diagram showing an example of the hardware configuration of the design support system shown in FIG.

[Embodiment 1]
Hereinafter, the design support system, the design support method, and the program according to the first embodiment of the present disclosure will be described with reference to the drawings. In the following description, for ease of understanding, the design object will be described as a semiconductor device including a plurality of circuit elements.

The design support system and method according to the present embodiment seek design constraints that relax the original design constraints to the extent that the final performance of the design object can be achieved without excessively exceeding the target value, and verify the design data. .. In the following description, for ease of understanding, the design constraint to be relaxed will be described as the external distance between circuit components. For the sake of distinction, the design constraint set at the design stage is called the first design constraint, and the relaxed design constraint is called the second design constraint within the range where the final performance can be expected to satisfy the target value. ..

FIG. 1 shows the configuration of the design support system 1 according to the first embodiment of the present disclosure. As shown in FIG. 1, the design support system 1 has a UI (User Threshold) device 10 that accepts and outputs information, and a second design constraint based on similar past design data that has been verified for product shipment. That is, a threshold value calculation device 20 for calculating the relaxed design constraint and a design verification device 30 for verifying whether or not the new design data 31 satisfies the design constraint are provided.

The design support system 1 satisfies the second design constraint, which is a relaxed design constraint, even when the external distance of the circuit element indicated by the new design data 31 does not satisfy the first design constraint, which is the original design constraint. In that case, the new design data 31 is verified as a pass.

The new design data 31 is created and supplied by the CAD (Computer Aided Design) device 40 for semiconductor device design.

FIG. 2 is a diagram illustrating the contents of the new design data 31. As shown in FIG. 2, the new design data 31 includes a semiconductor chip CH, data indicating a two-dimensional image and these positions of n circuit elements C _{1 ~} C _n arranged on the semiconductor chip CH Including. Note that n is an integer of 2 or more, FIG. 2 illustrates the case of n = 8, and i and j are natural numbers of 1 or more and n or less, which are different from each other. Further, the shapes of the circuit elements C ₁ to C _n will be described as a circle.

3, the radius _r i, _{r j} of circuit elements _{C i} and _{C j,} and the circuit element distance D _ij, defined as the distance between the center of the circuit elements _C i, _{C j,} circuitry _C i, C It is a figure exemplifying the _{outer shape distance di j} which is the distance between the outer shape of j. The circuit element distance D _ij , the radius r _i , r _j, and the external distance d _ij have a relationship of _{d ij} = D _ij − r _i − r _j. Further, the circuit element distance D _ij = D _ji and the external distance d _ij = d _ji are established. In the following, the external distance _dij will be described as a variable that defines design constraints, that is, a design variable. The design variable is a variable included in the design data for determining whether the design data satisfies the design constraint.

In this embodiment, "external distance d _ij ≥ reference distance d _thij " is set as the first design constraint. That is, it is set that the circuit elements C _i and C _j should be arranged at a distance of the reference distance d _{thij or more.} On the other hand, the design support system 1, the circuit elements C _i, outer shape distance d _ij of C _j, be less than the reference distance d _ThiJ, obtains a threshold d _Rthij which can be regarded as acceptable if higher, If the external distance d _{ij is equal to} or greater than the threshold d r th ij, it is judged as acceptable.

The UI device 10 shown in FIG. 1 has a display device that displays an image and is shown to the operator, a keyboard that accepts the operator's operation, a mouse, a tablet device, and a data input / output terminal such as a USB terminal.

The threshold value calculation device 20 corresponds to the design reference DB (database) 21 that stores the first design constraint, the past design DB 22 that stores the design data that has been verified for product shipment, and the relaxed design constraint. _{It includes} a threshold value calculation unit 23 for calculating the threshold value d rthij of the external distance d _ij , and a threshold value DB 24 for storing the _{calculated threshold value d r thij.}

The design reference DB 21 stores the design constraint set by the designer, that is, the first design constraint. As described above, in this embodiment, the design constraint includes a reference distance d _ThiJ contour distance d _ij circuit elements C _i and C _j. If contour distance d _ij circuit elements C _i and C _j is the reference distance d _ThiJ above, with respect to the outer distance, so that meets the first design constraint. The reference distance d _thij is a value common to a plurality of designs. The design reference DB 21 functions as a design reference storage unit. The reference distance d _thij is an example of the first design constraint value.

The past design DB 22 stores the verified design data. Here, the verified design data is past design data that has been verified for product shipment. The past design DB 22 stores, as the verified design data, the design data determined to be "passed" in the past verification, that is, the design data that has been verified to have substantially no problem and has been commercialized. Whether or not the design data is "passed" is determined by the results of various performance evaluation tests on the prototype of the product obtained from the design data.

The threshold value calculation unit 23 obtains a second design constraint based on the past design data stored in the past design DB 22. In this example, for the external distance _{dij of the circuit elements C i} and C _{j , the threshold value d r thij,} which is smaller than the reference distance d _thij but is the minimum value of the external distance that can be recognized as passing, is obtained. _{The method for obtaining the} threshold value d rthij will be described later. _{The threshold value DB 24} stores the threshold value drthij calculated by the threshold value calculation unit 23. The threshold value _drthij is an example of the second design constraint value.

_{When the calculated threshold value d rthij} does not exist in the threshold value DB 24, the threshold value calculation unit 23 _{additionally stores the calculated threshold value d rthij,} and when it already exists, the existing threshold value d _rthij is newly calculated. _Replace with drthij.

The new design data 31 is formed by the CAD device 40 and is supplied to the design support system 1 via a network or the like. As illustrated in FIGS. 2 and 3, the new design data 31 is data that defines the shape, size, position, and the like of the circuit elements of the semiconductor device to be designed. The design verification device 30 _{reads the reference distance d thij} from the design reference DB 21 and reads the _{threshold d r thij} from the threshold DB 24 with respect to the external distance.

The design verification device 30 obtains the _{external distance dij} and the reference distance d _thij for all combinations of _{adjacent circuit elements C i} and C _j included in the new design data 31. The design verification device 30 _{compares the obtained external distance d ij} with the reference distance d _thij, and if the external distance d _ij ≥ the reference distance d _thij , the first design constraint is satisfied, and the design item is passed. Determine.

On the other hand, design verification device 30, when the outer distance d _ij <reference distance d _ThiJ, comparing the contour distance d _ij and the threshold d _Rthij, if contour distance d _ij ≧ threshold d _Rthij, first design constraint However, since the second design constraint is satisfied, the design item is determined to be acceptable.

On the other hand, design verification device 30, when the outer distance d _ij <threshold _d rthij ≦ reference distance d _ThiJ, contour distance d _ij is too small, the first design constraints does not satisfy the second design constraints, failed To determine.

The design verification device 30 writes the verification result in the verification result output file 32. When the design verification device 30 _{completes verification for all combinations of adjacent circuit elements C i} and C _j , the design verification device 30 outputs a verification result output file 32.

_{Next, with reference to FIGS. 4 and 5, a procedure for obtaining} the second design constraint, more accurately, the threshold value drthij, will be described. Threshold calculation unit 23, a frequency distribution profile distance d _ij is a design variable, by analyzing using statistical processing to set the threshold value d _Rthij is a value indicating the second design constraints.
First, the threshold value calculation unit 23 uses all combinations of _{two circuit elements C i} and C _j that can be determined to be adjacent to each other among the _{n circuit elements C 1} to C n indicated by the design data stored in the past design DB 22. Ask. In the example of FIG. 2, the threshold value calculation unit 23 is, for example, a circuit element group {(C ₁ , C ₂ ), (C ₃ , C ₄ ), ..., (C _i , C _j ), ...・ (C _n-1 , C _n )} is obtained.

Next, the threshold value calculation unit 23 uses circuit element groups {(C ₁ , C ₂ ), (C ₃ , C ₄ ), ..., (C _i , C _j ), ... (C _n-1 , The _{external distance di j i} is extracted for each combination of the circuit elements C i and C _{j included in C n)}.} The extracted external distance _dij is distributed, for example, as illustrated in FIG.

Further, the threshold value calculation unit 23 reads the _{reference distance d thij} for each combination of i and j from the design reference DB 21.

Next, the threshold value calculation unit 23 extracts only the extracted external distance d _{ij that} is less than the reference distance d _{th ij.} The extracted external distance _dij does not satisfy the first design constraint, but is a value that is considered normal for the semiconductor device as a whole and has been commercialized. Therefore, even if the external distance di _j is set as these values, it is highly likely that no major problem will occur.

Threshold calculation unit 23, then, as illustrated in FIG. 5, the extracted contour distance d _ij, - the reference distance d _ThiJ contour distance d _ij] on the horizontal axis, vertical axis discrete existence probability [rho _d Create a graph as. The threshold value calculation unit 23 further obtains a cumulative existence probability σ indicating a cumulative value of _{the discrete existence probability ρ d.}

_{FIG. 5 divides} the [reference distance d thij-external distance d _ij ] into a plurality of sections, and shows the probability that the _{[reference distance d thij} -external distance d _{ij] exists in each of the plurality of sections. d} and the cumulative existence probability σ are illustrated. In FIG. 5, for a better distribution of discrete existence probability [rho _d, discrete existence probability [rho _d - shown in the reference distance d _ThiJ contour distance d _ij] interval graph format for each of the. Further, FIG. 5 does not necessarily correspond to FIG.

Since the actual existence probability ρ _d is continuous, the cumulative existence probability σ is shown by a continuous dotted line according to the distribution of the _{actual existence probability ρ d.} In other words, the threshold value calculation unit 23 _{statistically processes the external distance dij} included in the past design data, and as shown in FIG. 5, in each of the plurality of sections of _{[reference distance d thij} -external distance _dij]. The discrete existence probability ρ _d and the cumulative existence probability σ are calculated.

The threshold value calculation unit 23 obtains the _{external distance dij} corresponding to the portion where the cumulative existence probability σ reaches a predetermined set value P _{cdr of} the design support system 1, for example, 99%. The set value P _cdr is appropriately specified by the manufacturer or the operator using, for example, the UI device 10. The threshold value calculation unit 23 sets the obtained external distance d _ij as the threshold value d _{r th} ij. In FIG. 5, the cumulative existence probability σ up to the tenth section of _{[reference distance d thij} -external distance d _{ij ] is the maximum within a predetermined set value P cdr} , for example, 99%. It is a section. Therefore, the threshold value calculation unit 23 sets the external distance _dij corresponding to this point in the tenth section as the design threshold value _drth i j.

_{When the obtained threshold value d rthij is smaller than the conventional design threshold value d rthij} stored in the threshold value DB 24, the threshold value calculation unit 23 stores the obtained threshold value d _rthij as a new design threshold value d rthij in the threshold value DB 24, and stores the obtained threshold _{value d rthij in the threshold value DB 24.} To update. Alternatively, when the design threshold value d _{rthij of the combination of the circuit elements C i} and C _j is not stored in the threshold value DB 24, the threshold value calculation unit 23 sets the _{obtained design threshold value d r thij} as the combination of the circuit elements C _i and C _j. It _{is newly stored as the threshold value drthij} and the threshold value file is updated.

In this way, the threshold value calculation unit 23, a plurality of validated design data to market, variable extracting means for extracting the variable d _ij to determine the design constraints, the existence probability [rho _d of the extracted variable d _ij existence probability obtaining means for obtaining the cumulative presence probability obtaining means for obtaining a cumulative existence probability σ existence probability [rho _d determined, the value of the variable d _ij when the cumulative existence probability σ matches the preset reference value P _cdr particular However, it functions as a setting means for setting this value as a design constraint value. The external reference distance d _thij and the design threshold d _{r thij} are examples of design constraint values.

Next, the operation of the design support system 1 will be described.
The user operates the UI 10 and sequentially accumulates the past design data that has been verified and reached the commercialization in the past design DB 22.

The user takes in the new design data 31 to be verified. Further, the user incorporates the design constraint of the new design data 31 to be verified into the design reference DB 21.
Further, the user specifies, for example, the set value P _cdr and the number n of circuit elements included in the new design data 31 by using the UI device 10.

Next, the user instructs the verification of the new design data 31 by using the UI device 10.
In response to this instruction, the design support system 1 executes the threshold value calculation process shown in FIG. 6, and subsequently executes the verification process shown in FIG. 7.

Hereinafter, the threshold value calculation process and the verification process will be described with reference to FIGS. 6 and 7.
First, the threshold value calculation unit 23 sets the variables i and j to 1 when the threshold value calculation process shown in FIG. 6 is started (steps S11 and 12).

Next, the threshold value calculation unit 23 reads the external shape distance d _ij between circuit elements _{C i} and _{C j} of the designed circuit from the past design DB22 in the past (step S13). The threshold calculating unit 23, from the design reference DB 21, reads out the reference distance d _ThiJ contour distance between the circuit elements _{C i} and _{C j} (step S14).

Next, among the read external distances d _ij , those having a smaller reference distance d _th ij are extracted (step S15).
Threshold calculation unit 23, the extracted contour distance d _ij - calculating the reference distance d _ThiJ contour distance d _ij] (step S16).

Then, the threshold calculating unit 23, calculated - delimit the range [reference distance d _ThiJ contour distance d _ij] possible value into a plurality of sections. Next, the threshold value calculation unit 23 _{counts the number of [reference distance d thij} -external distance _{di ij} ] existing in each section, and as shown in FIG. 5, the discrete existence probability ρ _{d indicating the probability with respect to the whole.} (Step S17).

The threshold value calculation unit 23 _{accumulates the discrete existence probabilities ρ d} and obtains the cumulative existence probability σ (step S18).

Threshold calculation unit 23, when the cumulative existence probability σ matches the setting value _{P cdr} - determining the reference distance d _ThiJ contour distance d _ij] (step S19). Then, the threshold calculating unit 23, obtained - sought contour distance d _ij from the reference distance d _ThiJ contour distance d _ij] and the reference distance d _ThiJ, the outer distance d _ij obtained as a threshold value d _Rthij, threshold it Register in DB24 (step S19).

Next, the threshold value calculation unit 23 determines whether or not the variable j has reached n (step S20), and if not, sets j = j + 1 (step S21), proceeds to step S13, and continues the process. Incidentally, outline distance circuitry _{C i} and _{C j} contour distance d _ij and the circuit element C _j and C _i are equal. That is, _dij = _dji . Therefore, when updating j, the variable j is updated so that the processed _{dij is not processed again.}

When it is determined in step S20 that j = n (step S20: Yes), the threshold value calculation unit 23 determines whether or not the variable i has reached n (step S22), and if not (step S22:). No), i = i + 1 (step S21), the process proceeds to step S12, and the process is continued. As mentioned above, d _ij = d _ji . Therefore, when _{updating i, the variable i is updated so that the processed dij} is not processed again or is not equal to the variable j.

In this way, the external distances d ₁₂ , d ₁₃ , ... .. .. d _1n , d ₂₃ , d ₂₄ . .. .. d _2n , d ₃₄ , d ₃₅ ,. .. .. After _{processing d n-1, n} , it is determined in step S22 that j = n-1, and the process proceeds to the verification process of FIG. 7.

When the verification process is started, the design verification device 30 first sets the variables i and j to 1 (steps S31 and 32).

Then, the threshold calculating unit 23, the outer distance d _ij, whether they meet the first design constraints registered in the design criteria DB 21, i.e., whether the contour distance d _ij ≧ reference distance d _ThiJ Is determined (step S33). When the first design constraint is satisfied, that is, _{when it is determined that the external distance d ij} ≥ the reference distance d thij (step S33: Yes), the design verification device 30 determines that the design item is acceptable and verifies it. The result is registered in the verification result output file 32 (step S34).

_{On the other hand, when} it is determined that the first design constraint is not satisfied, that is, the external distance _dij <reference distance d thij (step S33: Yes), the external distance _dij is attached and registered in the threshold value DB 24. It is determined whether or not the design constraint of 2 is satisfied, that is, whether or not the external distance d _ij ≧ threshold value _{dr thij} is satisfied (step S35). When the second constraint is satisfied, that is, _{when it is determined that the external distance d ij} ≥ the threshold value _{dr thij} (step S36: Yes), the design verification device 30 determines that the design item has passed and determines the verification result. Register in the verification result output file 32 (step S34).

Further, when the second design constraint is not satisfied, that is, _{when it is determined that the external distance dij} <threshold value drthij (step S36: No), the design verification device 30 determines that the design item has failed. Then, the verification result is registered in the verification result output file 32 (step S37).

When it is determined in step S37 that j = n (step S37: Yes), the design verification device 30 determines whether or not the variable j has reached n (step S38), and if not, j = j + 1. (Step S38), the process proceeds to step S33, and the process is continued. As mentioned above, d _ij = d _ji . Therefore, when updating j, the variable j is updated so that the processed _{dij is not processed again and i and j do not match.}

When it is determined in step S38 that j = n (step S38: Yes), the design verification device 30 determines whether or not the variable i has reached n (step S38), and if not, i = i + 1. (Step S39), the process proceeds to step S32, and the process is continued. As mentioned above, d _ij = d _ji . Therefore, when updating i, the variable i is updated so that the processed _{dij is not processed again.}

In this way, the external distances d ₁₂ , d ₁₃ , ... .. .. d _1n , d ₂₃ , d ₂₄ . .. .. d _2n , d ₃₄ , d ₃₅ ,. .. .. Verification proceeds in order for d _{n-1, n.}
Finally, the design verification device 30 outputs the verification result output file 32 (step S40), and ends the process.

As described above, the threshold value calculation device 20 according to the present embodiment _{extracts the external distance dij} as a design variable from the verified design data, and statistically processes the frequency distribution of the extracted external distance _{dij. By analyzing using, the threshold value drthij} , which is a value indicating the second design constraint, is set. Thereby, an economical and rational second design constraint, that is, a threshold value can be automatically obtained within a range in which the final performance of the design object is expected to satisfy the target. As a result, it is possible to easily obtain a second design constraint, which is an appropriate relaxed design constraint with less personality. Therefore, it is possible to easily generate economical and rational design constraints in which the performance such as the electrical performance of the design product does not excessively exceed the target value.

Further, in the above embodiment, the threshold value is obtained from the design data for which the pass / fail judgment has been performed. Therefore, when checking the design constraint, it is possible to prevent the occurrence of a pseudo error corresponding to an error different from the pass / fail judgment given by the designer. To be more specific, such a pseudo-error is likely to occur when the outer distance d _ij is between the design threshold d _Rthij the design constraint value d _ThiJ, also design constraint value d _ThiJ design threshold It tends to occur when the difference with _{drthij is large.} In the design support system 1, each and every contour distance d _ij is, because it is compared with the design threshold d _Rthij corresponding to the outer shape distance d _ij, no pseudo check failures that can occur in visual check of the design drawing ..

Therefore, according to the design support system 1, circuit elements and the like as components constituting the device can be appropriately arranged. Further, as a result, the size of the device such as the size of the semiconductor chip CH can be reduced, and the device can be economically manufactured.

In the above embodiment, this disclosure has been described by taking as an example the design constraint that "the external distance d is equal to _{or greater than the reference distance d th".} This disclosure is not limited to this. For example, the design constraint that "the external distance d is equal to or less than the _{reference distance d th" can be similarly applied.}

In this case, as illustrated in FIGS. 8 and 9, for the external distance d _{ij larger than the reference distance d thij} , [external distance d _ij -reference distance d _thij ], its discrete existence probability ρ _d , and cumulative existence. The probability σ is calculated. The contour distance d _ij when the cumulative existence probability σ matches the reference value P _cdr is set as the threshold value _d rthij.

As a result, as shown in FIG. 8, when the external distance d _ij ≤ reference distance d _thij , the design item satisfies the first design constraint and passes, and the reference distance d _thij <external distance d _ij ≤ threshold value d. _{When rthij} , the design item does not satisfy the first design constraint, but it satisfies the second design constraint, so it passes. When the threshold value d _rthij <external distance _dij , the design item is the first design constraint. However, since the second design constraint is not satisfied, it is rejected.

In this embodiment, "external distance" is used as an example of variables that determine design constraints. The present embodiment can be similarly applied to any other variable, for example, design constraints such as inter-element distance D, radius r, and the like. Further, although this disclosure has been described by taking a circuit element of a semiconductor device as an example, the same can be applied to a circuit element of an arbitrary electric / electronic circuit and a mechanical element of a mechanical device. The same applies to the following embodiments.

In the present embodiment, in order to facilitate understanding, in order to obtain the existence probability and the cumulative existence probability of the variable that does not satisfy the first design constraint, the difference between the value of the variable and the design reference value is obtained and the difference is obtained. The discrete existence probability and the cumulative existence probability of were obtained. This disclosure is not limited to this method. A method of extracting variables that do not satisfy the first design constraint but are judged to pass as a whole and obtaining the existence probability and the cumulative existence probability of the extracted variables is arbitrary. For example, the extracted variables may be processed as they are without taking a difference. The same applies to the following embodiments.

[Embodiment 2]
In the first embodiment, the disclosure of the value of the variable has been described by taking as an example the design constraint that the value is equal to or greater than the reference value or equal to or less than the reference threshold value.
This disclosure is not limited to this. For example, a design constraint may be given in the form of "an expression having a plurality of variables is established". The processing in such a case will be described as the second embodiment.
The configuration of the design support system according to the present embodiment is the same as the configuration of the design support system 1 of the first embodiment shown in FIG. Further, the design data will be described as being the same as the design data shown in FIGS.

Here, design constraints, indicated with reference to Figure 3 - and a _{"d ij = D ij (r i} + r j) that is established." In addition, it is assumed that each variable is subject to an individual design constraint that it is equal to or higher than the reference value.

In this case, the threshold value calculation unit 23 specifies the circuit element distance D _ij , the radius r _i , the radius r _j , and the external distance _di j as variables of the formula constituting the design constraint.

These variables show distributions, for example, as illustrated in FIGS. 10A-10D.
In this embodiment, the circuit element distance D _ij, as shown in FIG. 10A, as satisfying the first design constraint when the circuitry distance D _ij ≧ reference value D _ThiJ, is determined to be acceptable, the reference value D _ThiJ > When the circuit element distance D _ij ≥ the threshold value Dr _thij , it is determined that the second design constraint is satisfied, and when the threshold value D _thij > the circuit element distance D _ij , it is determined to be rejected.

Further, as shown in FIG. 10B, the radius r _i is determined to be acceptable on the assumption that the first design constraint is satisfied when the _{radius r i} ≥ the reference value r _{thi , and the reference value r thi} > the radius r _i ≥ the threshold value r _thi. When the second design constraint is satisfied, it is determined to pass, and when the threshold value r _thi > radius r _i , it is determined to be unacceptable.

Similarly, as shown in FIG. 10C, the radius r _{j is determined to be acceptable on} the assumption that the first design constraint is satisfied when the _{radius r j} ≧ reference value r _{thj, and the reference value r thj} > radius r _i ≧ threshold value r. _{When thj} , it is determined that the second design constraint is satisfied, and when the threshold value r _thj > radius r _j , it is determined to be unacceptable.

Further, as shown in FIG. 10D, the external distance d _ij is determined to be acceptable as satisfying the first design constraint when the _{external distance d ij} ≥ the reference value d _{thij , and the standard value d thij} > the external distance d _ij ≧. When the threshold value d _thij is satisfied, it is determined that the second design constraint is satisfied, and when the threshold value Dr _thij > the circuit element distance D _ij , it is determined to be rejected.

When the threshold value calculation unit 23 obtains the threshold value D _{rthij of the circuit element distance D ij} , i) processes the design data stored in the past design DB ₂₂ (reference value D thij> circuit element distance D _ij ). The circuit element distance D _ij is extracted, and the discrete existence probability ρ _D of ii) (reference value D _thij -circuit element distance D _ij ) is obtained, and iii) the discrete existence probability ρ _D is accumulated and accumulated. The curve of the probability σ _D is obtained, and iii) the circuit element distance D _ij when the cumulative existence probability σ _{D equal to the predetermined value P Dcdr is given is set as the threshold value D rthij} of the circuit element distance and stored in the threshold value DB 24.

Further, when the threshold value calculation unit 23 obtains the threshold _{radius r ri of the radius r i} , i) processes the design data stored in the past design DB ₂₂ and (reference value r thri> radius r _i ). extract the radius r _i, ii) (a reference value r _ti - asking discrete existence probability [rho _ri radius r _i), iii) the curve of the cumulative presence probability sigma _ri by accumulating the discrete existence probability [rho _{ri Obtained} , iii) The radius r _i when giving the cumulative existence probability σ _ri equal to the predetermined value Pridr is set as the threshold r _ri of the radius r _i and stored in the threshold DB 24.

Similarly, the threshold value calculation unit 23, when determining the threshold value r _rj radius r _j is i) processing the design data stored in the past design DB 22, in (a reference value _r thrj> radius r _j) Extract a certain radius r _j , find the discrete existence probability ρ _rj of ii) (reference value r _tj -radius r _j ), and iii) _{accumulate the discrete existence probability ρ rj} and curve the cumulative existence probability σ _rj. look, the radius r _j when giving a cumulative existence probability sigma _rj equals iii) a predetermined value P _Rjcdr, as a threshold value r _rj radius r _j, and stores the threshold DB 24.

The threshold calculating unit 23, processing for obtaining the threshold value d _rij contour distance d _ij is the same as in the first embodiment.

However, the values of the integral values ∫ρ _D , ∫ρ _ri , ∫ρ _rj , and ∫ρ _d over the entire range of the existence probabilities ρ _D , ρ _ri , ρ _rj , and ρ _d are calculated to be equal to 1. .. That is, ∫ρ _D = ∫ρ _ri = ∫ρ _rj = ∫ρ _d = 1.

Design verification apparatus 30, when the i) circuitry distance D _ij ≧ threshold D _Rthij, determines that the circuit element distance D _ij to satisfy the design constraints, when ii) a radius r _i ≧ threshold r _ri, the radius r _It is determined that i satisfies the design constraint, and iii) when the radius r _j ≥ threshold r _rj , it _{is determined that the radius r j} satisfies the design constraint, and iv) the external distance d _ij ≥ threshold d _{r thij.} It _{is determined that the distance radius} satisfies the design constraint.

[Embodiment 3]
Hereinafter, the third embodiment of the present disclosure will be described with reference to FIG. 11, which shows the processing in the design support system 1 according to the embodiment. The configuration of the design support system of the third embodiment is the same as the configuration of the design support system 1 of the first embodiment. Further, also in the present embodiment, the first design constraint will be described as _{assuming that the external distance d ij} ≥ the reference value d th ij.

In FIG. 11A, the existence probabilities ρ _d of [reference value d _thij -external distance d _{ij ] of the circuit elements C i} and C _j included in the past design data are a plurality of extreme values, for example, two maximum values La and Lb. The case including is illustrated. 11B and 11C show the _{portions ρ da} and ρ _db including one of the maximum values La and Lb in the curve showing the _{existence probability ρ d} shown in FIG. 11A, respectively. _{11D and 11E show the threshold values} d arthij and d _br thij obtained from _{the existence probabilities ρ da} and ρ _db shown in FIGS. 11B and 11C, respectively.

As shown in FIG. 11A, a plurality of, for example, two maximum values La and Lb may occur in the existence probability ρ _d of [d _thij − _{di ij ] of the circuit elements C i} and C _{j included in the past design data.} .. Such a plurality of maximum values occur, for example, when the circuit elements C _i and C _j are commonly used in a semiconductor device housed in a plurality of types of semiconductor device packages having unique attributes. Packages for semiconductor devices having such a shape having a plurality of attributes include SOP (Small Outline Package) and BGA (Ball Grid Array).

The maximum value of each package attributes, the threshold calculating unit 23, a statistical distribution of the existence probability [rho _d of _[d thij -d _ij] Past Design DB22 circuitry C _i in the stored design _data, C _j It is obtained by analyzing in. The threshold value calculation unit 23 can obtain the attributes of the semiconductor device package and the like shown by the maximum values La and Lb shown in FIG. 11A by statistically analyzing the distribution of the _{existence probability ρ d.}

Is this case, the threshold value calculation unit 23, past designs DB22 on the stored circuit components _C i indicated design data, the existence probability [rho _d of _[d thij -d _ij] of _{C j,} a plurality of local maximum values are present Judge whether or not. Threshold calculation unit 23, two maximum values La of the presence probability [rho _d as shown in FIG. 11A, the curve of the existence probability [rho _d when the Lb is present, the partial curve [rho _da including a maximum value La as shown in FIG. 11B , Divided into a _{partial curve ρ db} including the maximum value Lb shown in FIG. 11C.

Then, the threshold calculating unit 23, only in the partial curve [rho _da including a maximum value La, sets the outline distance d _ij giving cumulative existence probability σ equal to a predetermined value P _cdr as the threshold value _d arthij. The threshold calculating unit 23, only in the partial curve [rho _db including a maximum value Lb, it sets the outline distance d _ij giving cumulative existence probability σ equal to a predetermined value P _cdr as the threshold value _d brthij.

_{When the existence probability ρ d} has three or more maximum values, the threshold value calculation unit 23 performs the same processing on each of the partial curves including the maximum values, and the threshold _values d arthij, d _brthij , d _crthij ... To get. The threshold value calculation unit 23 outputs a plurality of threshold _values d arthij, d _brthij , d _crthij ... Obtained by the above processing to the verification result output file 32. The predetermined value P _cdr may be common to a plurality of subcurves of the cumulative existence probability σ or may be different for each subcurve.

According to the design support system 1 according to the third embodiment, the circuit elements C _i, whether external distance d _ij between the C _j is appropriate to verify for each attribute, such as a package of a semiconductor device The accuracy of the verification result indicated by the verification result output file 32 can be improved.

[Embodiment 4]
Hereinafter, the fourth embodiment of the present disclosure will be described with reference to FIG. 12, which shows the processing in the design support system 1 according to this embodiment.

FIG. 12 is a diagram illustrating the contents of the new design data 31. The new design data 31 includes the semiconductor chip CH of the semiconductor device to be processed in the design support system 1 and two types of circuit elements C _a1 to C _aK and C _b1 to C _{bL arranged on the semiconductor chip CH.} Data indicating a dimensional image is included. In FIG. 12, K = 5, L = 3, k is an arbitrary number of 1 or more and K or less, and l is an arbitrary number of 1 or more and L or less.

Since semiconductor devices differ in the size and density of circuit elements, they may be designed by being divided into a plurality of regions AA and AB that can be represented by closed figures. For example, in the semiconductor device shown in FIG. 12, for example, a digital circuit is arranged in the area AA, and an analog circuit, for example, is arranged in the area AB.

Design verification device 30, the circuit elements _C a1 _{~ C aK} of area occupancy _{D A} in the region AA, is calculated by the following equation (2). However, in the following equation (2), _{S a} is the total sum of the areas of the circuit elements _C a1 _{~ C aK, S A} is the area of the entire area AA including a _{S a.}

_{D A = S a / S A} = ΣS ak / S A (2)

Design verification apparatus 30, the area occupancy _{D B} of the circuit element _C b1 _{~ C bL} in the region AB, is calculated by the following equation (3). However, in the following equation (3), _{S b} is the sum of the area of the circuit element _C b1 _{~ C bL, S B} is the area of the entire area AB containing _{S b.}

_{D B = S b / S B} = ΣS bl / S B (3)

Design verification apparatus 30, the area occupancy _D A, compares the value of _{D B,} when the D A _{<D B,} the value of the design threshold d _Athij the storage area AA of the threshold DB24 is, the area AB _Adjust the value of the design threshold d Brthij so that it is smaller than the value. The design verification device 30 is an example of the design constraint value adjusting means.

And more generalized, for example, when m number of regions 1 ~ m are provided in the semiconductor chip CH, design verification apparatus 30, the area occupancy _D 1 ~ _{D m} of the m regions _A 1 ~ _{A m} A table is created in which the calculated area occupancy rates D ₁ to D _m are arranged in descending order of value, for example, D ₁ <D ₂ <D ₃ << ... <D _m. Here, m is an integer of 2 or more.

In this case, design verification device 30, the value of the reference value _d 1thij ~ d mthij for regions _A 1 ~ _{A m} stored in the threshold DB 24, a _{d 1thij <d 2thij <d 3thij} <··· d mthij To change.
Design verification device 30 is thus changed reference value _d 1thij ~ d mthij, using as a threshold _d r1thij ~ d rmthij, circuit elements C _i included in each of the regions _{A 1 ~ A m ', C} j' It is determined whether or not the external distance _{di'j'between is appropriate.} However, i'and j'are 1 or more and are arbitrary numbers different from each other below the circuit elements included in each region.

In this way, when the semiconductor chip is divided into a plurality of regions, the occurrence of pseudo errors is reduced by changing the _{reference value d thij according to the area occupancy of the circuit element for each region and setting a threshold value.} Can be done.

[Embodiment 5]
Hereinafter, the fifth embodiment of the present disclosure will be described in detail with reference to FIG.
In the present embodiment, the threshold value DB 24 includes _{threshold value files 26 1} to 26 _{p associated with each of p attributes 1 to p.}

In addition to the analog circuit and digital circuit described above, the attributes of the semiconductor device include, for example, an analog / digital mixed circuit. The threshold file ₂₆ 1 ~ 26 _p respectively included in the threshold DB 24, the type of semiconductor device, or contains the value of threshold _d r1thij ~ d rpthij of each region of the semiconductor device.

When the operator performs an operation of designating the attributes of the semiconductor device or the attributes of each of the plurality of regions in the semiconductor device via the UI device 10, the design verification device 30 causes the threshold file 26 corresponding to the designated attributes. Read ₁ to 26 _p.

The design verification device 30 uses any of the threshold files 26 ₁ to 26 _p corresponding to the specified attribute to determine whether or not the external distance _dij between the circuit elements C _i and C _{j is appropriate.} to decide. Or, design verification apparatus 30 uses the respective threshold file 26 1 _~ 26 _p 2 or more corresponding to the attributes specified in the respective plurality of regions of the semiconductor device, circuit elements C _i included in each of the plurality of regions _', C _j' contour distance d _i'j between _'determines whether it is appropriate.

In this way, the occurrence of pseudo-check errors can be reduced by using _{the semiconductor device or the threshold value drthij} suitable for the attributes of each of the plurality of regions of the semiconductor device.

[Embodiment 6]
Next, the sixth embodiment will be described. In the above-described embodiments 1 to 5, the design support system 1 sets the design constraint based on the existence probability ρ _{d of the variable di j extracted from the design data.} On the other hand, in the sixth embodiment, the design support system 1 sets design constraints by using a machine learning method by AI (Artificial Intelligence). This will be described below.

The configuration of the design support system 1 according to the sixth embodiment is the same as the configuration shown in FIG.

The past design DB 22 stores the verified design data. Specifically, as shown in FIG. 14, the past design DB 22 stores a plurality of design data including design data A, design data B, design data C, ... As verified design data. Each of the design data A, the design data B, the design data C, ... Is data for designing the product of the model A, the model B, the model C, ..., For example. As an example of the design data A, B, C ..., Data for designing a mounting board in an electric / electronic device can be mentioned.

Further, as shown in FIG. 14, each of the design data A, B, C ... Stored in the past design DB 22 is labeled as "pass" or "fail" depending on the result of the past verification. There is. Past verification is performed, for example, by various performance evaluation tests on product prototypes obtained from design data. For example, the design data of the mounting board a judged to be acceptable in the performance evaluation for electrical / electronic equipment is labeled as "passed", and the design data of the mounting board b judged to be unacceptable is labeled Labeled "Fail".

Each of the design data A, B, C ... Stored in the past design DB 22 includes a plurality of variables that determine design constraints, as shown in FIG. Specifically, the design data A includes variables a1, a2, a3 ... That determine design constraints. Further, the variable a1 contains the variables a11, a12, a13 ..., The variable a2 contains the variables a21, a22, a23 ..., And the variable a3 contains the variables a31, a32, a33 ... The variables a11, a12, a13 ... Included in the variable a1 are, for example, the lengths of various elements included in the design data, and the variables a21, a22, a23 ... Included in the variable a2 are included in the design data, for example. The variables a31, a32, a33 ... Included in the variable a3 are, for example, the thicknesses between the various elements included in the design data. Like the design data A, the design data B, C ... Also include a plurality of variables that determine design constraints.

More specifically, when the design data A, B, C ... Are mounting boards in electrical and electronic equipment, as an example of variables, the distance to the bypass capacitor for each IC (Integrated Circuit) package such as BGA, SOP ... Can be mentioned. As described above, the design data A, B, C ... Stored in the past design DB 22 include a plurality of types of variables such as length, interval, thickness, distance, etc. as variables that determine design constraints. ..

Next, the threshold value calculation process executed by the threshold value calculation device 20 according to the sixth embodiment will be described with reference to the flowchart shown in FIG.

The threshold value calculation unit 23 extracts variables that determine design constraints from the verified design data A, B, C ... Stored in the past design DB 22 (step S51). As described above, the design data A, B, C ... Contain the variables a1, a2, a3 ... Of a plurality of types such as length and interval. The threshold value calculation unit 23 extracts a plurality of types of variables a1, a2, a3 ... Included in the design data A, B, C ... The threshold value calculation unit 23 is an example of the variable extraction means.

When variables are extracted from the design data A, B, C ..., the threshold value calculation unit 23 classifies the extracted variables into a plurality of groups using a clustering technique (step S52). After classifying the variables extracted from the design data A, B, C ..., each classification includes a plurality of types, so that the design constraints to be satisfied by all the variables are not always the same, and generally Must meet different design constraints. The threshold value calculation unit 23 classifies each variable for each corresponding design constraint as preprocessing in order to set the design constraint corresponding to each variable included in the design data A, B, C ....

Here, if it is not known which design constraint each variable should satisfy, it is generally difficult for the user to manually classify each variable, and it takes time and effort. Therefore, the threshold value calculation unit 23 classifies each variable into a plurality of groups by using a clustering technique which is a kind of unsupervised machine learning. The threshold value calculation unit 23 is an example of the classification means.

The threshold value calculation unit 23 uses the k-means method (k-means method) as a clustering technique. Specifically, the threshold value calculation unit 23 classifies each variable into k groups by executing the following processes (1) to (4).
(1) Classify each variable into k groups as appropriate. The value of k is preset by the user. For example, enter a number that is subject to conventional design constraints as the initial value of k. That is, the sum of the number of variables used by all design constraints is input as the initial value of k.
(2) Calculate the center of gravity of each group. As the parameter for calculating the center of gravity, information such as the size of the variable and the position coordinates can be used.
(3) Calculate the distance between the center of gravity of each group and each variable, and reclassify each variable into the group corresponding to the center of gravity closest to the distance.
(4) The processes of (2) and (3) are repeated until the change of the center of gravity of each group becomes equal to or less than a predetermined value.

As a result of classification by the threshold value calculation unit 23, variables of the same type may be classified into two or more different groups. In other words, since multiple design constraints may use variables of the same type, the types of variables after classification are not limited to all different among a plurality of groups, and may be duplicated among a plurality of groups. .. For example, the variables corresponding to the length may be classified into each of a plurality of different groups, and the variables corresponding to the interval may be classified into each of a plurality of different groups. Further, one group may include a plurality of types of variables. For example, the variables corresponding to the length and the variables corresponding to the interval may be classified into the same group.

When the variables are classified into a plurality of groups, the threshold value calculation unit 23 generates a frequency distribution of the variables for each group (step S53). Specifically, the threshold value calculation unit 23 generates a frequency table as shown in FIG. As an example, FIG. 17 shows a case where the variables extracted from the design data A, B, C ... Are classified into a group of the variable a1 and a group of the variable a2.

The threshold value calculation unit 23 generates a frequency distribution of variables included in each group. Specifically, the threshold value calculation unit 23 takes each value among the plurality of variables included in the group of the variable a1, such as one that takes 1 mm, two times that takes 2 mm, and so on. Aggregate the frequency of variables. Further, the threshold value calculation unit 23 similarly totals the frequencies of the variables that take each value for other groups such as the group of the variable a2. In this way, the threshold value calculation unit 23 generates a frequency table as shown in FIG. 17 by aggregating the frequencies of the variables included in each of the plurality of groups.

The threshold value calculation unit 23 sets such a frequency distribution table as a variable extracted from the design data labeled "pass" and a variable extracted from the design data labeled "fail". Generate for each of. As a result, the threshold value calculation unit 23 generates a frequency distribution determined to be "passed" and a frequency distribution determined to be "failed" for each of the plurality of groups.

Returning to the threshold value calculation process shown in FIG. 16, when the frequency distribution is generated, the threshold value calculation unit 23 calculates a threshold value indicating a design constraint for each group based on the generated frequency distribution (step S54). .. Specifically, the threshold value calculation unit 23 analyzes the generated frequency distribution for each of the plurality of groups using unsupervised machine learning. As a result, the threshold value calculation unit 23 sets the design constraints that the variables in each group should satisfy. In the example of the mounting board, the threshold calculation unit 23 divides the variable data group extracted from the design data determined to be "pass" and the variable data group extracted from the design data determined to be "fail". Based on this, the design constraint on the distance to the capacitor is calculated for each IC package mounted on the mounting board.

As an example, FIG. 19 shows a frequency distribution of "pass" and a frequency distribution of "fail" in a certain group. The threshold value calculation unit 23 searches for a boundary that separates the “pass” frequency distribution and the “fail” frequency distribution by unsupervised learning.

Specifically, the threshold value calculation unit 23 uses the initial value of the preset threshold value as a boundary, and sets the ratio of the variables existing on the pass side and the fail side in the frequency distribution of "pass" and "fail". The ratio of variables existing on the pass side and the fail side in the frequency distribution of "" is calculated. Then, the threshold value calculation unit 23 determines which of the ratio of the variables existing on the passing side in the "pass" frequency distribution and the ratio of the variables existing on the failing side in the "failing" frequency distribution. The threshold value is updated from the initial value so that the value becomes moderately large.

Here, an example of a threshold calculation method by unsupervised machine learning will be described. The threshold value calculation unit 23 sets a temporary threshold value d, and calculates a pass area Sp (d) and a fail area Sf (d) determined by the threshold value d. Then, the threshold calculation unit 23 calculates the average pass area Spave (d) obtained by averaging the pass area Sp (d) with the number of sample parameters by the following equation (1), and sets the fail area Sf (d) as the parameter. The average rejected area Sfave (d) averaged by the number of samples of is calculated by the following equation (2). In equations (1) and (2), n and m represent the number of parameter samples.

When the average pass area Shave (d) and the average fail area Sfave (d) are calculated, the threshold value calculation unit 23 calculates the evaluation function J (Spave (d), Sfave (d)) represented by these. The threshold value calculation unit 23 calculates the evaluation function J (Spave (d), Sfave (d)) by changing the value of the threshold value d in various ways, and the evaluation function J (Spave (d), Sfave (d)) is the maximum. The threshold value d is searched for. As a result of the search, the threshold value calculation unit 23 sets the threshold value d at which the evaluation function J (Spave (d), Sfave (d)) is maximized as the final threshold value.

The accuracy of the threshold value calculated by machine learning in this way increases as the number of design data A, B, C ... Stored in the past design DB 22 increases, that is, as the number of variables increases. Further, by using an existing threshold value that the user has conventionally recognized as an initial value, the accuracy of the calculated threshold value can be improved. The threshold value calculation unit 23 is an example of the setting means.

When the difference between the threshold value, which is the design constraint value obtained by machine learning, and the threshold value, which is the existing design constraint value, is larger than a predetermined assumed value, the threshold value calculation unit 23 uses the UI device 10 via the UI device 10. Output a warning. In other words, when the design constraint obtained by machine learning deviates from the range assumed for the existing design constraint, the threshold value calculation unit 23 notifies the user of a warning.

However, even if a warning is output, the user can use the design constraints obtained by machine learning. This is because even if the design constraints are outside the expected range, there is a possibility that the target performance can be obtained as long as it is obtained by machine learning. For example, the threshold value calculation unit 23 receives an instruction from the user via the UI device 10 as to whether or not to use the design constraint value for which the warning is output. When the instruction to use the design constraint value for which the warning is output is received, the design verification device 30 verifies the new design data 31 using the design constraint value.

Returning to the threshold value calculation process shown in FIG. 16, when the threshold value calculation unit 23 calculates the threshold value for each group, the calculated threshold value is stored in the threshold value DB 24 (step S55). Specifically, the threshold value calculation unit 23 generates the design constraint table shown in FIG. 18 and stores it in the threshold value DB 24. The design constraint table is a table in which the design constraints calculated for each of a plurality of groups are identifiablely stored for each group. In the example of FIG. 18, the design constraint table defines that the variable a1 is 4 mm or less as a design constraint in the group of the variable a1, and the variable a2 is 4 mm or more as a design constraint in the group of the variable a2. Has been established.

By generating the design constraint table in this way, the threshold value calculation unit 23 sets the design constraints in each of the plurality of groups. The threshold value DB 24 is an example of a design constraint storage unit that stores a plurality of design constraints.

As described above, the design support system 1 according to the sixth embodiment extracts variables that determine design constraints from the verified design data, and classifies the extracted variables into a plurality of groups using a clustering technique. Then, the design support system 1 according to the sixth embodiment sets a threshold value which is a value indicating a design constraint for each group by analyzing the frequency distribution of the variable in each of the plurality of groups by using machine learning. In this way, since design constraints are set using machine learning, procedural processing by mathematical modeling is not required. Therefore, even if the frequency distribution of variables has an unexpected shape and mathematical modeling is difficult, the performance of the design product does not exceed the target value excessively, which is an economical and rational design constraint. Can be easily set.

Further, in the sixth embodiment, the threshold value calculation unit 23 classifies all the variables extracted from the design data A, B, C ... into a plurality of groups by using a clustering technique. However, the threshold value calculation unit 23 may narrow down the variables to be classified from the variables extracted from the design data A, B, C ... Before the classification process. For example, the threshold value calculation unit 23 receives from the user the selection of the variable for determining the design constraint from the variables included in the design data A, B, C ... From the user via the UI device 10. Then, the threshold value calculation unit 23 may classify the received variables into a plurality of groups by using a clustering technique.

Further, in the sixth embodiment, the threshold value calculation unit 23 sets the threshold value as a design constraint based on the frequency distribution of "pass" and the frequency distribution of "fail". However, the threshold value calculation unit 23 may generate a frequency distribution of variables extracted only from the design data labeled "pass" and set design constraints based only on the "pass" frequency distribution. .. Specifically, the threshold value calculation unit 23 calculates the threshold value, which is a value indicating the design constraint to be satisfied by the variable, by analyzing the frequency distribution of “pass” by unsupervised machine learning. Even if the design constraint is set based only on the frequency distribution of "pass", the same quality effect can be obtained although the accuracy is lower than that based on the frequency distribution of "pass" and "fail".

[Embodiment 7]
Next, the seventh embodiment will be described. In the sixth embodiment described above, the threshold value calculation unit 23 classifies the variables extracted from the design data A, B, C ... into a plurality of groups by using a clustering technique. On the other hand, in the seventh embodiment, information on which of the plurality of design constraints the variables included in the design data A, B, C ... Are subject to is given in advance.

The configuration of the design support system 1 according to the seventh embodiment is the same as the configuration shown in FIG.

The past design DB 22 stores the verified design data A, B, C ... Labeled as "pass" or "fail", respectively, as in the sixth embodiment. In the seventh embodiment, the past design DB 22 further stores the correspondence between the variables and the design constraints as shown in FIG. 20 in advance. In the correspondence shown in FIG. 20, each variable such as length, interval, thickness, and distance included in the design data A, B, C ... is a design constraint among a plurality of design constraints P, Q, R ... It defines whether it is a target.

The threshold value calculation device 20 according to the seventh embodiment executes the same process as the threshold value calculation process in the sixth embodiment shown in FIG. However, since the design constraints corresponding to each variable are predetermined, the classification process in step S52 is different from that of the sixth embodiment.

Specifically, the threshold value calculation unit 23 extracts variables that determine design constraints from the verified design data A, B, C ... Stored in the past design DB 22 (step S51). When the variables are extracted, the threshold value calculation unit 23 classifies the extracted variables into a group of corresponding design constraints (step S52). Specifically, the threshold value calculation unit 23 refers to the correspondence between the variables stored in the past design DB and the design constraints, and classifies the extracted variables into a group of design constraints corresponding to the variables. ..

When the variables are classified, the threshold value calculation unit 23 generates the frequency distribution of the variables for each group (step S53). When the frequency distribution is generated, the threshold value calculation unit 23 calculates a threshold value indicating a design constraint for each group based on the generated frequency distribution (step S54). When the threshold value is calculated, the threshold value calculation unit 23 stores the calculated threshold value in the threshold value DB 24 (step S55). Since the processing of steps S53 to S55 is the same as that of the sixth embodiment, the description thereof will be omitted.

As described above, the design support system 1 according to the seventh embodiment extracts variables that determine design constraints from the verified design data, and divides the extracted variables into a plurality of groups prepared in advance for each design constraint. Classify. Then, the design support system 1 according to the seventh embodiment sets a threshold value, which is a design constraint, for each group by analyzing the frequency distribution of variables in each of the plurality of groups using machine learning. Since the correspondence between the variables and the design constraints is predetermined, in addition to the effect of the sixth embodiment, the effect that the design constraints can be set even for the variables that cannot be classified by the clustering technique can be obtained.

[Embodiment 8]
Next, the eighth embodiment will be described. The design support system 1 according to the eighth embodiment performs design verification of new design data by using the design constraints set by the threshold value calculation device 20 according to the sixth and seventh embodiments.

The configuration of the design support system 1 according to the eighth embodiment is the same as the configuration shown in FIG. The threshold value DB 24 stores the design constraint table generated by the threshold value calculation device 20 according to the sixth and seventh embodiments.

The verification process executed by the design verification device 30 according to the eighth embodiment will be described with reference to the flowchart shown in FIG.

When the verification process is started, the design verification device 30 accepts the input of the verification condition, which is the condition for performing the design verification (step S71). Specifically, the design verification device 30 selects the design constraint to be applied to the new design data 31 from the plurality of design constraints stored in the threshold value DB 24 as a verification condition via the UI device 10. Accept from the user. Further, the design verification device 30 accepts the selection of the type and variables of the new design data 31 to be verified as the verification condition from the user via the UI device 10. The UI device 10 is an example of a reception means.

Upon receiving the input of the verification condition, the design verification device 30 verifies the new design data 31 (step S72). Specifically, the design verification device 30 extracts the variable to be verified received in step S71 from the new design data 31. Then, the design verification device 30 refers to the design constraint table stored in the threshold value DB 24, and verifies whether or not the extracted variable satisfies the design constraint of the application target accepted in step S71. The design verification device 30 is an example of the design verification means.

When the new design data 31 is verified, the design verification device 30 outputs the verification result (step S73). Specifically, the design verification device 30 writes output information indicating whether or not each variable included in the new design data 31 satisfies the design constraint in the verification result output file 32 shown in FIG. 22.

In the example of FIG. 22, the variable a11 existing at the coordinates (x1, y1) satisfies the design constraint because its value is 4 mm or less. Therefore, the design verification device 30 determines that the variable a11 has passed. On the other hand, the variable a12 existing at the coordinates (x2, y2) does not satisfy the design constraint because its value is 4 mm or more. Therefore, the design verification device 30 determines that the variable a12 has failed.

The design verification device 30 outputs output information indicating whether or not each variable included in the new design data 31 satisfies the design constraint to the outside by displaying it on the display of the UI device 10. The UI device 10 is an example of an output means for outputting output information.

As described above, the design support system 1 according to the eighth embodiment verifies the new design data 31 by using the design constraints set automatically with high accuracy by machine learning. Thereby, it is possible to verify whether or not the performance of the designed product satisfies the target value.

In the eighth embodiment, the design verification device 30 verifies the new design data 31 using the design constraints set by the machine learning method by the threshold value calculation device 20 according to the sixth and seventh embodiments. However, the design verification device 30 may be able to switch between the design constraint set by machine learning and the existing design constraint set without machine learning. .. For example, the design verification device 30 uses either a design constraint set by machine learning or an existing design constraint set without machine learning for each variable included in the new design data 31. The user may accept the selection of whether to perform the verification as one of the verification conditions in step S71.

Further, the design verification device 30 is not limited to using the design constraint set by the threshold value calculation device 20 according to the sixth and seventh embodiments as it is, and is a new design in which the design constraint set by the threshold value calculation device 20 is corrected. The newly designed data 31 may be verified by using the constraint. For example, the design verification device 30 may verify the new design data 31 by using a new threshold value obtained by multiplying the threshold value set by the threshold value calculation device 20 by the safety factor. The safety factor may be predetermined or may be accepted from the user as one of the verification conditions in step S71.

Further, the threshold value calculation unit 23 may update the design constraint by Bayesian update based on the verification result of whether the variable included in the new design data 31 satisfied with the design constraint, which is verified by the design verification device 30. .. Specifically, the threshold value calculation unit 23 adds the frequency of the variable extracted from the newly designed data 31 and determined to be "pass" by the verification to the frequency distribution of "pass" already used for machine learning. , Update the frequency distribution of "pass". The threshold value calculation unit 23 also updates the frequency distribution of "failure" in the same manner. Then, the threshold value calculation unit 23 sets a new design constraint by analyzing the updated frequency distributions of "pass" and "fail" using unsupervised machine learning. By sequentially updating the design constraints by Bayesian update in this way, the design constraints can be set with higher accuracy.

Further, the threshold value calculation unit 23 may accept corrections to the verification result verified by the design verification device 30 from the user, and correct the design constraint based on the verification result to which the received corrections have been added. Specifically, the user inputs the correction of the verification result of each variable output to the verification result output file 32 via the UI device 10. For example, the user corrects the verification result of the variable determined to be acceptable in the verification by the design verification device 30 to be unacceptable by his / her own judgment. Alternatively, the user corrects the verification result of the variable determined to be unsuccessful in the verification by the design verification device 30 to be acceptable by his / her own judgment. When the correction of the verification result is accepted, the threshold value calculation unit 23 adds the "pass" and "fail" variables in the corrected verification result to the frequency distributions of "pass" and "fail" already used for machine learning, respectively. The frequency distribution of "pass" and "fail" is updated by adding the frequency of. Then, the threshold value calculation unit 23 sets a new design constraint by analyzing the updated frequency distributions of "pass" and "fail" using unsupervised machine learning. As a result, design constraints can be set more flexibly in consideration of the user's judgment.

FIG. 23 is a diagram illustrating a hardware configuration of a computer 6 that realizes the design support system 1 according to the above-described embodiments 1 to 8. As shown in FIG. 23, the computer 6 includes a CPU (Central Processing Unit) 302 that executes a program instruction code, a memory 304 including a RAM (Random Access Memory), a ROM (Read Only Memory), and an HDD (Hard Disk Drive). ), A storage device 306 including an SSD (Solid State Drive), a display device, a mouse, and the like, and an input / output device 308 that can be used as the UI device 10 are connected via a bus 300.

Each component of the design support system 1 shown in FIG. 1 is realized by a program running on the computer 6. The distribution method of such a program is arbitrary, and is stored in a computer-readable recording medium such as a CD-ROM (CompactDiskROM), DVD (DigitalVersatileDisk), MO (MagnetoOpticalDisk), or memory card. It may be distributed via a communication network such as the Internet.

Further, each component of the design support system 1 is not limited to the CPU 302, the memory 304, etc. shown in FIG. 23, and is executed by, for example, a digital circuit, an analog circuit, an FPGA (Field Programmable Gate Array), a microcontroller, or the like. It can be realized by combining with a program. The CPU 302, digital circuit, analog circuit, FPGA, etc. that operate each component of the design support system 1 can be collectively referred to as a control unit or a processor.

Further, the first to eighth embodiments can be arbitrarily combined as long as they do not cause mutual contradiction. Further, in the first to fifth embodiments, the case where the design support system 1 verifies the design data of the semiconductor device is illustrated, but the design support system 1 is the design data of an arbitrary electric circuit, electronic circuit, or mechanical device. Used for verification.

Further, in the first to eighth embodiments, the case of verifying the two-dimensional design data is illustrated, but the design support system 1 is used for generating the threshold data of the three-dimensional design data and verifying the design items. Can be done. Further, although a circle is illustrated as the shape of the circuit element, the shape of the component is arbitrary.

Further, when the circuit element C _i, the shape of the C _j other than circle, the point used to define and circuitry distance D _ij and contour distance d _ij between them, for example, is defined as those centers It is the point, the position of the center of gravity of these, and so on. The reference point may be set appropriately because it is afraid of the shape of the circuit element.

In the first to fifth embodiments, the threshold value calculation unit 23 extracts a variable that does not satisfy the first design constraint from the verified design data, and based on the existence probability of the extracted variable, from the first design constraint. Also set a relaxed second design constraint. However, the first design constraint may not be provided. When the first design constraint is not provided, the threshold value calculation unit 23 extracts a variable that defines the design constraint from the verified design data regardless of whether or not the first design constraint is satisfied. Specifically, the threshold value calculation unit 23 extracts the _{external distance di j} for all combinations of _{two adjacent circuit elements C i} and C j. Then, the threshold value calculation unit 23 may set a design constraint based on the existence probability of the extracted variable. If the first design constraint is not provided, the design support system 1 may not include the design reference DB 21 that stores the first design constraint. The same applies to the following embodiments.

On the other hand, in the sixth and seventh embodiments, the threshold value calculation unit 23 extracts all the variables that determine the design constraint from the design data A, B, C ... Regardless of whether or not the first design constraint is satisfied. did. However, the threshold value calculation unit 23 extracts variables that do not satisfy the first design constraint from the design data A, B, C ..., as in the first to fifth embodiments, and is based on the frequency distribution of the extracted variables. Therefore, a second design constraint that is relaxed from the first design constraint may be set.

The design reference DB 21, the past design DB 22 and the threshold value DB 24 are not limited to being provided inside the design support system 1, but may be provided outside the design support system 1. For example, each DB may be provided in a data server that provides resources in cloud computing. In this case, the design support system 1 writes data to each DB and reads data from each DB by communicating with a data server via a wide area network such as the Internet.

The present disclosure allows for various embodiments and modifications without departing from the broad spirit and scope of the present disclosure. Moreover, the above-described embodiment is for explaining this disclosure, and does not limit the scope of the present disclosure. That is, the scope of the present disclosure is indicated by the scope of claims, not by the embodiment. And various modifications made within the scope of the claims and within the equivalent meaning of disclosure are considered to be within the scope of this disclosure.

This application is based on the international application PCT / JP2019 / 034386 filed on September 2, 2019. The specification, claims, and drawings of the international application PCT / JP2019 / 034386 are incorporated herein by reference.

1 design support system, 3 computer, 300 bus, 302 CPU, 304 memory, 306 storage device, 308 input / output device, 10 UI device, 20 threshold calculation device, 21 design standard DB, 22 past design DB, 23 threshold calculation unit, 24 threshold DB, 26 threshold file, 30 design verification device, 31 new design data, 32 verification result output file, 40 CAD device.

Claims (20)

1. Variable extraction means for extracting variables that determine design constraints from verified design data,
   A setting means for setting the design constraint by analyzing the frequency distribution of the variable extracted by the variable extraction means by using statistical processing or machine learning.
   Design support system equipped with.
2. Existence probability acquisition means for obtaining the existence probability of the extracted variable, and
   Further provided with a cumulative existence probability acquisition means for obtaining the cumulative existence probability of the obtained existence probability.
   The setting means identifies the value of the variable when the cumulative existence probability matches a preset reference value, and sets this value as the design constraint value of the design constraint.
   The design support system according to claim 1.
3. The variable extraction means extracts a plurality of variables that determine the design constraint, and obtains a plurality of variables.
   The existence probability acquisition means obtains the existence probability of each of the extracted plurality of variables, and obtains the existence probability.
   The cumulative existence probability acquisition means obtains the cumulative existence probability of each existence probability of the plurality of variables, and obtains the cumulative existence probability.
   The setting means sets each value of the variable when the cumulative existence probability matches a preset reference value as the design constraint value of the design constraint.
   The design support system according to claim 2.
4. The existence probability acquisition means includes means for obtaining one or more new existence probabilities based on attributes from the obtained existence probabilities.
   The cumulative existence probability acquisition means obtains the cumulative existence probability of a new existence probability.
   The design support system according to claim 2 or 3.
5. When the obtained existence probability has a plurality of extreme values, the existence probability acquisition means obtains a plurality of new existence probabilities so as to separate the plurality of extreme values.
   The cumulative existence probability acquisition means obtains the cumulative existence probability of each of the plurality of new existence probabilities.
   The setting means identifies the value of the variable when each cumulative existence probability matches a preset reference value, and sets this value as the design constraint value of the design constraint.
   The design support system according to any one of claims 2 to 4.
6. The variable extraction means extracts a variable that does not satisfy the first design constraint set at the design stage as a variable that determines the design constraint from the verified design data.
   The setting means obtains a second design constraint that is relaxed from the first design constraint by analyzing the frequency distribution of the variable extracted by the variable extraction means by using statistical processing or machine learning. Set,
   The design support system according to any one of claims 1 to 5.
7. Further provided with a design reference storage unit for storing the first design constraint that specifies that the value of the variable is equal to or less than or equal to the first design constraint value.
   The variable extraction means extracts the variable having a value less than or larger than the first design constraint value from the verified design data.
   The design support system according to claim 6.
8. A design constraint value adjusting means for obtaining the occupancy rate of a plurality of parts in the area of the device to be designed and adjusting the design constraint value according to the occupancy rate of the plurality of the parts is provided.
   The design support system according to any one of claims 1 to 7.
9. The occupancy rate of a plurality of parts in the area of the device to be designed is obtained for each of the plurality of areas of the device, and the occupancy rates of the plurality of the parts in each of the plurality of the areas are determined. A design constraint value adjusting means for adjusting the design constraint value of the above is provided.
   The design support system according to any one of claims 1 to 8.
10. The design constraint value adjusting means adjusts the design constraint value to a larger value as the occupancy rate of the plurality of parts increases.
    The design support system according to claim 8 or 9.
11. Further provided with a classification means for classifying the variables extracted from the verified design data into a plurality of groups.
    The setting means sets the design constraint for each group based on the frequency distribution for each group of the variables classified by the classification means.
    The design support system according to any one of claims 1 to 10.
12. The classification means classifies the variables extracted from the verified design data into the plurality of groups by using a clustering technique.
    The design support system according to claim 11.
13. The setting means sets the design constraint by analyzing the frequency distribution of the variable extracted by the variable extraction means using unsupervised machine learning.
    The design support system according to any one of claims 1 to 12.
14. The variable extraction means extracts the variable from each of the design data determined to be acceptable by the verification and the design data determined to be unacceptable by the verification.
    The setting means uses the unsupervised machine learning to obtain the frequency distribution of the variables extracted from the design data determined to be acceptable and the frequency distribution of the variables extracted from the design data determined to be unacceptable. By analyzing the above, the design constraint is set.
    The design support system according to claim 13.
15. Further provided with a design verification means for verifying whether or not the variables included in the new design data satisfy the design constraint set by the setting means.
    The design support system according to any one of claims 1 to 14.
16. Further provided with a receiving means for accepting the selection of the design constraint to be applied to the new design data from the plurality of design constraints.
    The design verification means verifies whether or not the variables included in the new design data satisfy the design constraints accepted by the reception means.
    The design support system according to claim 15.
17. The setting means updates the design constraint based on the verification result of whether or not the variable included in the new design data satisfies the design constraint, which is verified by the design verification means.
    The design support system according to claim 15 or 16.
18. The setting means receives a modification to the verification result from the user, and updates the design constraint based on the verification result to which the accepted modification is added.
    The design support system according to claim 17.
19. From the verified design data, the variables that define the design constraints are extracted and
    The design constraints are set by analyzing the frequency distribution of the extracted variables using statistical processing or machine learning.
    Design support method.
20. On the computer
    Processing to extract variables that determine design constraints from verified design data,
    A process of setting the design constraint by analyzing the frequency distribution of the extracted variable using instrumental processing or machine learning.
    A program that executes.

Images